Proactive Depression Screening via Human-Integrated AI: Synergizing Multimodal Smartphone Biomarkers and Game As Reality for the Global South

Introduction

Indonesia faces critical mental health challenges, with depression affecting an estimated 15 million people, yet a staggering treatment gap of over 90% persists due to a psychiatrist shortage and deeply rooted stigma. Traditional screenings require trained clinicians and have high friction due to systemic stigma, making it difficult to be scalable across a geographically dispersed archipelago. To bridge this critical structural gap, we introduce GAR (Game-As-Reality) Well, a human-integrated AI framework powered by the "Game As Reality" (GAR) approach. By transforming rigid clinical assessments into engaging gameful wellness interactions, it utilises everyday smartphones as accessible and proactive diagnostic instruments where the act of scanning constitutes an engaging and non-stigmatising experience that captures multimodal biomarkers.

Methods

We developed a multimodal architecture synergising photoplethysmography (PPG), voice biomarkers, and subjective symptoms from brief self-reports into a unified screening model. The gameful design ensures users perceive the process as a wellness check rather than a psychiatric assessment, addressing Indonesia's cultural stigma barrier. We conducted in-silico validation using 50,148 epidemiologically matched profiles, calibrated to Indonesian demographics with Monk Skin Tone Scale corrections to ensure inclusive and melanin-aware algorithmic fairness.

Results and Conclusions

The GAR-Well framework achieved diagnostic performance with a test ROC-AUC of 0.886 and an overall accuracy of 90.6%. Crucially, the model maintained equitable performance across all age, gender, and ethno-phenotypic sub-groups. By effectively synergising multimodal objective sensing with human clinical context through a "Game As Reality" interface, GAR-Well has the potential to overcome both the structural and psychological bottlenecks of traditional mental health screening. While current findings rely on rigorous in silico validation, they establish a definitive foundation for forthcoming real-world clinical trials. It has the potential to dismantle stigma and to democratise mental health detection in Indonesia and the Global South, directly advancing the core objectives of SDG 3.